Spaces:
Build error
Build error
Update app.py
Browse files
app.py
CHANGED
|
@@ -6,8 +6,7 @@ from backend.rag import retriever
|
|
| 6 |
from backend.llm_local import generate
|
| 7 |
from backend.summarizer import extractive_sents, map_reduce
|
| 8 |
|
| 9 |
-
|
| 10 |
-
|
| 11 |
settings.RAW_DIR.mkdir(parents=True, exist_ok=True)
|
| 12 |
settings.INDEX_DIR.mkdir(parents=True, exist_ok=True)
|
| 13 |
|
|
@@ -25,7 +24,7 @@ def upload_files(files):
|
|
| 25 |
dest = settings.RAW_DIR / f.name
|
| 26 |
dest.parent.mkdir(parents=True, exist_ok=True)
|
| 27 |
f.save(str(dest))
|
| 28 |
-
return gr.update(choices=list_uploaded_files(), value=None), "Upload
|
| 29 |
|
| 30 |
def build_index():
|
| 31 |
global STATE_INDEXED
|
|
@@ -34,14 +33,14 @@ def build_index():
|
|
| 34 |
retriever.build()
|
| 35 |
retriever.load()
|
| 36 |
STATE_INDEXED = True
|
| 37 |
-
return "Índice criado
|
| 38 |
|
| 39 |
def chat_answer(history, message):
|
| 40 |
if not STATE_INDEXED:
|
| 41 |
try:
|
| 42 |
retriever.load()
|
| 43 |
except Exception:
|
| 44 |
-
return history + [("system","
|
| 45 |
hits = retriever.search(message, top_k=settings.TOP_K)
|
| 46 |
ctx = hits[:settings.TOP_K_RERANK]
|
| 47 |
ans = generate(message, ctx)
|
|
@@ -53,24 +52,24 @@ def summarize_run(filename, pages, chapter, query, style, length):
|
|
| 53 |
try:
|
| 54 |
retriever.load()
|
| 55 |
except Exception:
|
| 56 |
-
return "
|
| 57 |
page_start = page_end = None
|
| 58 |
if pages:
|
| 59 |
try:
|
| 60 |
a,b = pages.split("-"); page_start, page_end = int(a), int(b)
|
| 61 |
except:
|
| 62 |
-
return "Formato de páginas inválido. Use
|
| 63 |
if chapter and filename:
|
| 64 |
rng = retriever.find_chapter_range(filename, chapter)
|
| 65 |
if not rng:
|
| 66 |
-
return "
|
| 67 |
page_start, page_end = rng["start"], rng["end"]
|
| 68 |
hits = retriever.search(query or "resumo", top_k=16,
|
| 69 |
filename=filename or None,
|
| 70 |
page_start=page_start, page_end=page_end)
|
| 71 |
if not hits:
|
| 72 |
-
return "
|
| 73 |
-
|
| 74 |
sents = extractive_sents(hits, query, max_sents=80)
|
| 75 |
if sents:
|
| 76 |
synth = [{"text": s, "meta": {"page_num": p}} for s,p in sents]
|
|
@@ -79,8 +78,8 @@ def summarize_run(filename, pages, chapter, query, style, length):
|
|
| 79 |
final = map_reduce(hits, focus=query or "resumo", style=style, length=length)
|
| 80 |
return final
|
| 81 |
|
| 82 |
-
with gr.Blocks(title="Agente de Estudos IA") as demo:
|
| 83 |
-
gr.Markdown("# Agente de Estudos IA
|
| 84 |
|
| 85 |
with gr.Tab("Upload & Indexar"):
|
| 86 |
files_in = gr.File(label="Envie PDFs / PPTX / TXT", file_count="multiple", type="filepath")
|
|
@@ -112,3 +111,6 @@ with gr.Blocks(title="Agente de Estudos IA") as demo:
|
|
| 112 |
|
| 113 |
run_btn.click(summarize_run,
|
| 114 |
inputs=[file_dd, pages_tb, chapter_tb, query_tb, style_dd, length_dd],
|
|
|
|
|
|
|
|
|
|
|
|
| 6 |
from backend.llm_local import generate
|
| 7 |
from backend.summarizer import extractive_sents, map_reduce
|
| 8 |
|
| 9 |
+
# inicializa diretórios
|
|
|
|
| 10 |
settings.RAW_DIR.mkdir(parents=True, exist_ok=True)
|
| 11 |
settings.INDEX_DIR.mkdir(parents=True, exist_ok=True)
|
| 12 |
|
|
|
|
| 24 |
dest = settings.RAW_DIR / f.name
|
| 25 |
dest.parent.mkdir(parents=True, exist_ok=True)
|
| 26 |
f.save(str(dest))
|
| 27 |
+
return gr.update(choices=list_uploaded_files(), value=None), "✅ Upload feito! Clique em Indexar."
|
| 28 |
|
| 29 |
def build_index():
|
| 30 |
global STATE_INDEXED
|
|
|
|
| 33 |
retriever.build()
|
| 34 |
retriever.load()
|
| 35 |
STATE_INDEXED = True
|
| 36 |
+
return "✅ Índice criado/carregado. Vá para a aba Conversar ou Resumir."
|
| 37 |
|
| 38 |
def chat_answer(history, message):
|
| 39 |
if not STATE_INDEXED:
|
| 40 |
try:
|
| 41 |
retriever.load()
|
| 42 |
except Exception:
|
| 43 |
+
return history + [("system","⚠️ Monte o índice na aba Upload & Indexar")], ""
|
| 44 |
hits = retriever.search(message, top_k=settings.TOP_K)
|
| 45 |
ctx = hits[:settings.TOP_K_RERANK]
|
| 46 |
ans = generate(message, ctx)
|
|
|
|
| 52 |
try:
|
| 53 |
retriever.load()
|
| 54 |
except Exception:
|
| 55 |
+
return "⚠️ Monte o índice na aba Upload & Indexar."
|
| 56 |
page_start = page_end = None
|
| 57 |
if pages:
|
| 58 |
try:
|
| 59 |
a,b = pages.split("-"); page_start, page_end = int(a), int(b)
|
| 60 |
except:
|
| 61 |
+
return "Formato de páginas inválido. Use '12-30'."
|
| 62 |
if chapter and filename:
|
| 63 |
rng = retriever.find_chapter_range(filename, chapter)
|
| 64 |
if not rng:
|
| 65 |
+
return "Capítulo não encontrado no sumário (TOC)."
|
| 66 |
page_start, page_end = rng["start"], rng["end"]
|
| 67 |
hits = retriever.search(query or "resumo", top_k=16,
|
| 68 |
filename=filename or None,
|
| 69 |
page_start=page_start, page_end=page_end)
|
| 70 |
if not hits:
|
| 71 |
+
return "Nenhum conteúdo encontrado para o escopo."
|
| 72 |
+
# pequenas sentenças chaves + map-reduce
|
| 73 |
sents = extractive_sents(hits, query, max_sents=80)
|
| 74 |
if sents:
|
| 75 |
synth = [{"text": s, "meta": {"page_num": p}} for s,p in sents]
|
|
|
|
| 78 |
final = map_reduce(hits, focus=query or "resumo", style=style, length=length)
|
| 79 |
return final
|
| 80 |
|
| 81 |
+
with gr.Blocks(title="Agente de Estudos IA 📚") as demo:
|
| 82 |
+
gr.Markdown("# Agente de Estudos IA 📚\nCarregue seu **livro** e **slides** e estude com o bot.")
|
| 83 |
|
| 84 |
with gr.Tab("Upload & Indexar"):
|
| 85 |
files_in = gr.File(label="Envie PDFs / PPTX / TXT", file_count="multiple", type="filepath")
|
|
|
|
| 111 |
|
| 112 |
run_btn.click(summarize_run,
|
| 113 |
inputs=[file_dd, pages_tb, chapter_tb, query_tb, style_dd, length_dd],
|
| 114 |
+
outputs=out_md)
|
| 115 |
+
|
| 116 |
+
demo.queue().launch()
|